If you're going to use Sharp IR rangefinders, realize that they have a few important limitations:
1) minimum range (this will probably be the biggest problem)
2) narrow beam width (you could just mount it on a servo...)
3) nonlinear output
For detecting walls, a simpler/more reliable setup would be a set of ultrasonic rangefinders, because their field of view is usually plus/minus 20 degrees - enough to create a "bumper" system; and also because some of them have no minimum range limit (big plus). Since sonar isn't as cheap as IR, you could mount a single sensor on a panning servo and just turn the sensor to view the three sides (left-front-right).
For detecting the floor, you probably want an IR-LED/phototransistor pair, which will be able to tell you the color of the floor. Sharp IR rangefinders are essentially incapable of this because of the minimum range issue, and are much too expensive for this purpose anyways.
P.S.: Just a personal pet peeve - just because this is an internet forum doesn't mean you can avoid typing in english - "chat" or "IM" abbreviations are not standardized, and will confuse people - or make them think that you actually did not know how to spell the word (which I doubt, so it's probably intentional)